After using sftp:// with nautilus for a minute or two, attempting to enter a directory causes nautilus to lockup with 100% CPU usage. A dialog asking "Cancel Open?" (attached below) pops up. Clicking Cancel does not recover. Killing nautilus seems to be the only way to recover. Description of problem: poll([{fd=4, events=POLLIN}, {fd=3, events=POLLIN}, {fd=8, events=POLLIN|POLLPRI}, {fd=10, events=POLLIN}, {fd=12, events=POLLIN|POLLPRI}, {fd=16, events=POLLIN}, {fd=38, events=POLLIN}], 7, 0) = 0 ioctl(3, FIONREAD, [0]) = 0 gettimeofday({1096236610, 778163}, NULL) = 0 poll([{fd=4, events=POLLIN}, {fd=3, events=POLLIN}, {fd=8, events=POLLIN|POLLPRI}, {fd=10, events=POLLIN}, {fd=12, events=POLLIN|POLLPRI}, {fd=16, events=POLLIN}, {fd=38, events=POLLIN}], 7, 0) = 0 brk(0x8f96000) = 0x8f96000 brk(0x8f86000) = 0x8f86000 brk(0x8f74000) = 0x8f74000 brk(0x8f6e000) = 0x8f6e000 brk(0x8f6b000) = 0x8f6b000 ioctl(3, FIONREAD, [0]) = 0 gettimeofday({1096236610, 783878}, NULL) = 0 poll([{fd=4, events=POLLIN}, {fd=3, events=POLLIN}, {fd=8, events=POLLIN|POLLPRI}, {fd=10, events=POLLIN}, {fd=12, events=POLLIN|POLLPRI}, {fd=16, events=POLLIN}, {fd=38, events=POLLIN}], 7, 0) = 0 futex(0x896a1e4, FUTEX_WAKE, 1) = 1 futex(0x896a1e0, FUTEX_WAKE, 1) = 1 futex(0x896a1c0, FUTEX_WAKE, 1) = 1 ioctl(3, FIONREAD, [0]) = 0 gettimeofday({1096236610, 786155}, NULL) = 0 poll([{fd=4, events=POLLIN}, {fd=3, events=POLLIN}, {fd=8, events=POLLIN|POLLPRI}, {fd=10, events=POLLIN}, {fd=12, events=POLLIN|POLLPRI}, {fd=16, events=POLLIN}, {fd=38, events=POLLIN}], 7, 0) = 0 ioctl(3, FIONREAD, [0]) = 0 gettimeofday({1096236610, 786354}, NULL) = 0 poll([{fd=4, events=POLLIN}, {fd=3, events=POLLIN}, {fd=8, events=POLLIN|POLLPRI}, {fd=10, events=POLLIN}, {fd=12, events=POLLIN|POLLPRI}, {fd=16, events=POLLIN}, {fd=38, events=POLLIN}], 7, 0) = 0 brk(0x8f96000) = 0x8f96000 brk(0x8f86000) = 0x8f86000 brk(0x8f74000) = 0x8f74000 brk(0x8f6e000) = 0x8f6e000 brk(0x8f6b000) = 0x8f6b000 ioctl(3, FIONREAD, [0]) = 0 gettimeofday({1096236610, 792392}, NULL) = 0 Version-Release number of selected component (if applicable): nautilus-2.8.0-1
Created attachment 104338 [details] cancel_open.png
Need a stacktrace for all threads of nautilus when this happens, with nautilus and gnome-vfs2 debuginfo packages installed.
I was able to reproduce this readily while connected to my old RH box... either 7.3 or 8.0. Unfortunately I have since formatted that box. Tests connecting to FC3, FC2, RHEL3, RHEL2.1, Solaris with commerical sshd seem to not exhibit this problem. Should I keep this bug open?
Closing for now until I am able to reproduce it again.